Expansion and adaptation of the M₅B₁₂O₂₅(OH) structure type to incorporate di- and trivalent transition metal cations

2.1 Synthesis

The used starting materials were H₃BO₃ (≥99.8 %, Carl Roth, Karlsruhe, Germany) and B₂O₃ (99.9 %, Strem Chemicals, Newburyport, MA, USA) as boron sources and one of the respective oxides (V₂O₅ (≥99.6 %, Merck, Darmstadt, Germany), Cr₂O₃ (99 %, Merck, Darmstadt Germany), MnO (99 %, Sigma Aldrich, St. Louis, MO, USA), Mn₂O₃ (99 %, Sigma Aldrich, St. Louis, MO, USA) or Fe₂O₃ (≥99 %, Sigma-Aldrich, St. Louis, MO, USA)). To reduce V₂O₅, a stoichiometric quantity of elemental B (95 %, Goodfellow, Huntingdon, GB) replaced an equal amount of B₂O₃. To receive Cr₅B₁₂O₂₅(OH), some Cr₂O₃ was replaced by Cr(NO₃)₃·9 H₂O (99 %, Sigma Aldrich, St. Louis, MO, USA), and only B₂O₃ was used as a boron source; with increasing amount of oxide, larger crystals were obtained, however, the amount of CrBO₃ as a side product also increased. The starting materials were mixed in a stoichiometric fashion, thoroughly ground in an agate mortar and encapsulated in a Pt capsule and placed in an α-BN (Henze Boron Nitride Products AG) crucible with a fitting lid of the same material. All experiments were carried out in 18/11 assemblies and the syntheses were performed via HP/HT methods using a Walker-type multianvil module within a 1000 t hydraulic press (Max Voggenreiter GmbH). A two-step process consisting of six steel wedges and eight tungsten carbide cubes (Hawedia) as outer and inner anvils was used to build up the pressure. Further details concerning this kind of setup are described in the literature [7], [8], [9].

The pressure, at which the syntheses were carried out, was 9 GPa for all substances, except for the synthesis of V₅B₁₂O₂₅(OH), where 8 GPa were sufficient. Pressure was built up in 240 min, followed by heating to the desired temperature within 10 min. For V₅B₁₂O₂₅(OH) and Cr₅B₁₂O₂₅(OH), the maximum temperature was 1000 °C. The maximum temperature of the reaction was held for 60 min, followed by a step-wise reduction to 900 °C and eventually 800 °C with 2 K min⁻¹, with keeping the temperature constant for 60 min at 900 and 800 °C. The maximum temperature for Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7 and Mn₅Mn_0.83B₁₂O₂₆ was 800 °C. After 20 min at 800 °C, the temperature was decreased to 400 °C within 200 min. Mn₅MnB₁₂O₂₂(OH)₄ was heated to 1200 °C, with a lowering of the temperature to 800 °C in 200 min after 20 min holding time at maximum temperature. Afterwards, the reaction mixture was quenched to room temperature within a few minutes in all cases. Decompression to ambient conditions was performed within 720 min. The Pt capsules were cleaned before collecting the crystalline products. An overview of all syntheses is shown in Table 1.

2.2 Phase identification and structure determination by X-ray diffraction

The products of the syntheses were large crystalline blocks of varying color, which formed together with one to three by-products. The crystalline products were thoroughly ground, mounted on a flat sample holder and analyzed by X-ray diffraction on a STOE Stadi P powder diffractometer. Measurements were performed in transmission geometry with Ge(111)-monochromatized MoK-L₃ radiation (λ = 0.7093 Å) within a range of 2θ = 2–70°, a step size of 0.015° and a Mythen 1 K detector. The Topas 4.2 software [10] was used for the Rietveld refinement. One representative refinement is shown in Figure 3, the others are shown in the Supporting Information in Figures S1–S4.

Suitable crystals were isolated under a Leica 125M polarization microscope and measured with a Bruker D8 Quest diffractometer equipped with a Photon 300 CMOS detector. Multiscan absorption correction was performed with Sadabs-2016/2 [11]. The program suite WinGX-2018.1 [12] was used for the calculation of the structures, which were solved and refined with the implemented ShelXT-2018/2 [13] and ShelXL-2018/3 [14, 15] routines, respectively.

Further details of the crystal structure investigations may be obtained from the joint CCDC/FIZ Karlsruhe online deposition service: https://www.ccdc.cam.ac.uk/structures/? by quoting the deposition numbers CSD 2294261 for V₅B₁₂O₂₅OH, CSD 2294262 for Cr₅B₁₂O₂₅OH, CSD 2294263 for Mn₅MnB₁₂O₂₂(OH)₄, CSD 2294264 for Mn₅Mn_0.83B₁₂O₂₆, and CSD 2294265 for Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7.

2.3 Electron microprobe measurements

Single crystals of Mn₅Mn_0.83B₁₂O₂₆ and Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7 were investigated via electron microprobe measurements using a JEOL JXA-iSP100 Super Probe equipped with five WDS spectrometers and a fully integrated JEOL EDS system. The acceleration voltage used was 15 kV, with a beam current of 10 nA. To ensure electrical conductivity, a thin layer of carbon was deposited on the sample. The calibration samples used for the WDX measurement were magnetite (Fe₃O₄) and rhodonite (MnSiO₃). K lines were used for the quantitative measurements.

3.1 Experimental adaptations to unexpected findings

Initially, the aim of the conducted syntheses were the substances M₅B₁₂O₂₅(OH) (M = V³⁺, Cr³⁺, Mn³⁺, Fe³⁺) from V₂O₅, Cr₂O₃/Cr(NO)₃, Mn₂O₃, and Fe₂O₃ in combination with B₂O₃ and H₃BO₃. This targeted synthesis worked well for V₅B₁₂O₂₅(OH) and Cr₅B₁₂O₂₅(OH). However, the crystals of what was meant to be Mn₅B₁₂O₂₅(OH) and Fe₅B₁₂O₂₅(OH) immediately showed two issues that needed to be addressed: their cell volumes were too large when compared with the ionic radii of Mn³⁺ and Fe³⁺, and the significant residual electron density in the cuboctahedral cavities within the interpenetrating borate networks.

To investigate these unexpected findings, we conducted bond valence sum (BVS) calculations and performed electron microprobe measurements to rule out any contamination of the product with other elements and to detect a possible difference in composition.

One evidence is based on the M2–O distances, which are considerably longer than the corresponding M1–O distances, which results in a larger M2O₆/M1O₆ ratio (see Table 6). This is a strong indication that the octahedrally coordinated M2 position is occupied with a divalent metal cation instead of a trivalent one. On the other hand, the qualitative EDX measurements showed no other element than the respective metal, boron, and oxygen in the crystals, which means that the large cuboctahedral cavities have to be occupied by the comparably small transition metal cations. Based on these findings, we performed another synthesis with MnO as starting material This synthesis yielded the compound Mn₅MnB₁₂O₂₂(OH)₄, in which the cuboctahedral cavity is completely filled with Mn²⁺. The crystals are colorless in agreement with the oxidation state +2, while the crystals are brown when synthesized from Mn₂O₃.

The solid solution series with the end members M^II₅M^IIB₁₂O₂₂(OH)₄ and M^III₅B₁₂O₂₅(OH) therefore represent the occupation of M sites with divalent and trivalent cations, respectively. The charge compensation of the anionic borate framework B₁₂O₂₆H _x ^(16−x)− might vary, depending on the incorporated elements and their specific properties.

3.2 Powder X-ray diffraction and Rietveld refinements

Figure 4 shows a representative Rietveld refinement of Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7. The other Rietveld refinements are shown in the Supporting Information available online in Figures S1–S4. All occurring side products are listed in Table 2.

Figure 4:

Rietveld refinement of Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7. The signal at 2θ = 8.2°, marked with an asterisk, stems from grease used to prepare a flat sample.

3.3 Description of the products

Using the reaction conditions for the syntheses described above, most substances form large crystalline blocks of distinct color. This makes the separation from any side products relatively simple, except for Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7, for which two black side products are observed, which are known in the literature. Table 2 lists the color of the product and the side products of the syntheses; Figure 5 shows photographs of the samples. Some side products have not been known yet in the literature and will be reported elsewhere.

Figure 5:

Photographs of the synthesized borates and their respective side products (Table 2). P = product, SP = side product.

3.4 Crystal structures

The five transition metal borates all crystallize in the space group I4₁/acd (no. 141), like all the known tetragonal representatives of this structure type [1], [2], [3]. One unit cell contains eight formula units (Z = 8). The lattice parameters vary between a = 1105.7–1178.98 pm and c = 2159.59–2217.46 pm, with a total cell volume V ranging from 2.6783 to 3.0823 nm³. In relation to their ionic radii, the cell volumes are in accordance with those of the published substances containing the interpenetrating (B₁₂O₂₆)¹⁶⁻ networks, with the exception of Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7, which shows a larger cell volume than the ionic radius of iron would suggest. The two substances with the largest cell volumes, Mn₅MnB₁₂O₂₅(OH) and Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7, exhibit a split position for the M2 cations. This phenomenon is also known in the literature for Ga₅B₁₂O₂₅(OH) and In₅B₁₂O₂₅(OH) [2]. The three substances that (partly) incorporate a cation in the cuboctahedral cavity, i.e. Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7, Mn₅MnB₁₂O₂₂(OH)₄, and Mn₅Mn_0.83B₁₂O₂₆, show a slightly larger a axis, as seen in Figure 6. Mn₅MnB₁₂O₂₂(OH)₄ and Mn₅Mn_0.83B₁₂O₂₆ also show a distinctly shortened c axis. This is due to the fact, that all three compounds feature the larger divalent cation rather than the trivalent cation on the M2 position. As the cuboctahedral cavity shows larger M3–O distances than the sum of the ionic radii of the respective cation and oxygen, the best structural refinements were achieved with a split position of the cation residing on M3.

Figure 6:

Comparison of the cationic radii and the lattice parameters of the title compounds. Hollow dots are known from the literature, the full dots are from this publication. The trend line is calculated only from the literature-known values. Substances with different cations on the positions M1 and M2 are written as M1/M2.

A list containing details of the structure refinements based on the single-crystal data is found in Table 3. Bond valence sums (BVS) were calculated for all substances with the bond-length/bond-strength concept [21, 22] and the charge distributions using the charge distribution in solids (CHARDI) concept [23, 24] are listed in Table 4. The interatomic distances are given in Table 5. Lists containing the atomic coordinates, thermal displacement factors and bond angles are given in the Supporting Information in Tables S1–S15.

The discrepancy of the calculated BVS with the formal oxidation numbers for the M2 position is due to the longer M2–O distances compared to M1–O distances, which are very close the formal oxidation number, as both positions contain the same cation and are octahedrally coordinated. This variation on size of the octahedra is intrinsic to this structure type; hence, most BVS calculations yield a lower charge for the M2 position, when both positions are occupied with the same cation. One exception is Ga₄InB₁₂O₂₅(OH), as in this case the appropriately larger cation occupies the larger octahedron [3]. For all substances with mixed valency, we therefore conclude that the cation with the higher oxidation number is found in the M1 octahedra, in contrast to the first publication on this field, which stated that Ti⁴⁺ was on the M2 position [1].

To compare the different substances, we analyzed the volume of the coordination polyhedra. It can be seen that the volume of the cuboctahedron relative to that of the octahedra decreases with the increasing occupancy of the central cation position. This makes sense, as the cation reduces the Coulomb repulsion between the vertices of the polyhedron. Furthermore, the volume of the M2O₆ octahedra is larger when occupied with a divalent rather than a trivalent cation. These values and trends can be observed in Table 6. Based on the size of the Fe1O₆ octahedron and the ”higher than expected” cell volume we conclude that it hosts both: the larger Fe²⁺ and smaller Fe³⁺ cations, either in different domains of the single crystal or as a part of a solid solution series between M₅B₁₂O₂₅(OH) and M₅MB₁₂O₂₂(OH)₄.

3.5 Electron microprobe measurements

To rule out the contamination with another element, single crystals of Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7 and Mn₅Mn_0.83B₁₂O₂₆ were analyzed quantitatively with WDX and qualitatively and semiquantitatively with EDX. No other elements than B, O, and the respective metal could be detected. The standards used for calibration were natural minerals: Fe₃O₄ and MnSiO₃. Due to the lack of reference materials for B and O, these elements were added as fixed, based on the expected composition of the sample to ensure proper matrix correction. Representative EDX spectra for both substances are depicted in Figure 7. Figure 8 shows BSE pictures of the samples, with an overview in (a) and (c) and a focus on crystals, which appeared to be single-crystalline in (b) and (d). In fact, several crystal domains can be seen; the difference in color indicates a variation in the metal to borate ratio. These ratios all fall within the solid solution series M₆B₁₂O₂₂(OH)₄ – M₅B₁₂O₂₅(OH) when factoring in the error of the measurements, which are the combination of using different calibration compounds (up to ∼1 % w/w) and the standard deviation of results. Hence, these results offer an explanation for the partial occupation of the cuboctahedral cavities. Table 7 lists the averaged results of the quantitative WDX measurements, as well as the minimum and maximum value measured for the specific metal and the respective standard deviation of the results.

Figure 7:

EDX spectra of Mn₅Mn_0.83B₁₂O₂₆ (top) and Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7 (bottom).

Figure 8:

BSE pictures of Mn₅Mn_0.83B₁₂O₂₆ (a, b) and Fe₅Fe_0.14B₁₂O_24.3(OH)_1.7 (c, d).

We hence conclude that the crystals in the systems Mn–B–O(–H) and Fe–B–O–H in fact show several domains of the solid solution series between completely divalent and trivalent metal cations. In the case of the iron compound, the protonation and occupation of the cuboctahedral M3O₁₂ site within the FBBs plays a role, whereas the needed charge compensation in the manganese containing compound is primarily achieved by filling the M3O₁₂ site with the metal cation.